Share of voice is the most quoted number in communications reporting and the least interrogated. It appears on every dashboard because it is easy to produce: count the mentions, divide by the category total, present the percentage.
That number is frequently wrong in the only way that matters. It tells you that you were mentioned. It does not tell you whether being mentioned helped.
Volume is not position
An unweighted share of voice count treats a considered feature in a business daily and a passing reference in an aggregator repost as equivalent events. It treats a critical mention and a supportive one as equivalent. Run that measure for four quarters and it will reliably reward the organisation making the most noise, not the one holding the strongest position.
We have seen this produce genuinely perverse incentives. A team hits its share of voice target during a quarter in which sentiment deteriorated, because the volume driving the number came from coverage they would not have chosen. The metric went up. The position got worse. Nobody was lying.
Three corrections
Weight by publication authority. Not all placements carry the same weight with the audiences that decide procurement, policy, or investment. Build a tiered list before the campaign starts, not after the results come in. Deciding what counts once you can see the scores is not measurement.
Weight by sentiment, and separate neutral from absent. A neutral mention is not a win. It is frequently the most honest signal available: you were present in the conversation and gave the reader no reason to change their mind.
Measure against a named competitor set. Category-wide share of voice flatters organisations in fragmented categories and punishes them in concentrated ones. The useful question is narrower: against the three organisations you actually lose to, is your position improving?
What to report instead
Report weighted share of voice against a named set, with sentiment split out, and a plain sentence on what moved and why. If the number rose because of one high-volume story you did not place and would not repeat, say so.
A metric that cannot go down in a bad quarter is not measuring anything. It is decoration, and this firm has a view about decoration.